JOURN A L OF RESEARCH of the Nationa l Bureau of Standards - A. Physics and Chemistry

Vol. 70A, No. 1, January-Februa ry 1966

Phase Equilibria in the System Niobium Pentoxide-­Boric Acid

Ernest M. Levin

(October 11 , 1965)

The phase-equilibrium diagram for the system Nb20 5 -B20 3 was de termined experime ntally using the quenching technique and examining the samples by optical r:nicroscopy and. x -r~r powder diffractom etry. The system contaInS one bInary compound of approxImate composIt Ion 3Nb,Os' 8,03," which melts incongruently at about 1150 0 e to H - Nb20 5 and B20 3' rich liquid. A large region of liquid immiscibility extends, at 13~2 °e, from 10 mole percent Nb,Os to 65. 7 mole perce nt Nb,Os. The theoreti cally calculated ex te nt of ImmIscIbIlIty IS In reasonable agreement wIth the expe rImentally d etermined value.

Key Words : Boric oxide, immiscibility, niobium pe ntoxide, oxides, phase equilibri a .

1. Introduction

Consis tent with the objective expressed in the Intro­duction to the preceding paper on the system Nb20 5 -

Ge02 [1],1 the systematic study of immiscibility was extended to include the phase relations in the sys te m Nb20 5 - B20 3 • The Nb5+ cation has a high charge and moderate ionic radius, 0_69 A; and, consequently, a high ionic field strength. Notwithstanding the high field strength, the Nb20 5 - Ge02 system does not show liquid immiscibility_ However, the Nb20 5 - Si02

system was found to contain a large region of immis­cibility [2]. Based on the observation that no case has been reported in which liquid immiscibility exists for the silicate system but not for the corresponding borate system, immiscibility in the Nb20 5 -B20 3

system was to be expected. Furthermore, the system offered a possible opportunity to apply and to exte nd the principles of immiscibility developed mainly from systems of borates with the oxides of divalent cations.

2. Sample Preparation and Test Methods

Starting materials for the preparation of mixtures consisted of high-purity niobium pentoxide and re­agent grade boric acid powder. The Nb20 5 contained the following impurities when examined by the general qualitative spectrochemical method: Si -less than 0.1 percent; Fe, Sn, Ti-O.OOI to 0.01 percent; Ca, Mg-0.0001 to 0_001 pe rcent; Cu - ? The boric acid was of especially high purity, containing only faint traces (0.0001 to 0.001%) of Mg and Si.

Three-gram batches, on an ignited basis, of the starting materials were formulated by mixing or grind­ing, pressing, and heating for 4 hr, at three succ;s­sively higher temperatures: 700,800, and about 900 C. Phase equilibrium relations were studied by the well­established quenching technique. Samples were heated in sealed Pt tubes for periods of 1 to 336 hr.

t Figures in bracke ts indicate the lite rature refe rences at the end of thi s paper .

1 1

Te mperatures were meas ured with a calibrated Pt-90Pt: 10 Rh thermocouple . Quenc hed sa mples were examined with the binoc ular and polarizing mi cro­sco pes and by x-ray powde r diffractometry (Ni-filtered CuKa radiation) usin g a high-angle Geiger-counter diffrac tometer. The technique and details of sample preparation, apparatus, and method have been de­scribed in previou s publications [3] and we re the same as for the study of the Nb20 5 -Ge02 sys te m [1]. The overall maximum uncertainty of the liquidus te mpera­tures re ported are es timated to be within ± 10 °C.

It s hould be noted that the polarizing microscope was of limited value because clear homogeneous o-lass was not observed, even for the co m positions richest in B20 3. Que nched liquid devitrifi ed, to lesse r or greater extent, and the grain s appeared brownish, finely mottled, and very weakly birefringe nt, whe n viewed under the mi croscope.

3. Chemical Analysis

Formulated compositions were analyzed by the Analytical Chemistry Division. For most of th e com­positions only the Nb20 5 content was determined, by titration of an unbuffed solution at pH 4.5 with stand­ard EDT A solution. Percent B20 3 was es timated by difference from 100 perce nt. Analyses of three com­positions (formulated at 25, 50, and 66.7 mole percent Nb20 5) as well as analysis of the mechanically sepa­rated heavy and light devitrified liquids in the immi s­cibility region were made for both Nb20 5 and B20 3 •

B20 3 was determined by titration of a neutral solution (PH 7.00) with standard NaOH solution, after the addi­tions of manitol. Nb20 5 was determined from the weight of the ignited residue. Analyzed values were used in constructing the phase diagram. Those com­positions for which one of the co m pone nts was deter­mined by difference have an estimated uncertainty of about 2 mole percent. Those for which both compo­nents were analyzed have an estimated accuracy of better than 1 mole percent.

1400

()

I 1

0 1

()

1 65.7 r'

1300 () () () () ()

u o ~ 1200 w 0:: :::> r-

() ()

()

()

()

13 () ()

<{ 0::

• • • 13 () . . () ()

W

~ 1100 w r-

H-Nb205 + 1000 "3Nb205·S2021

I I I I I

900 • • • I() I I I I I I

()

2-LlQUIDS

o

()

()

()

()

() ()

o

()

()

()

()

() ()

() ()

, , I ,

10.0 I ~

()

()\ , I

()I \

\ I I

() , , ()

() ()

800~~~~~~~~~--~~~~~~~~--~~~

Nb205 10 20 30 40 50 60 70 80 90

------- -

13NbO:BO" 2 5 2 3

MOL.%

FIGURE 1. Phase diagram for the system Nb20 S - B20 3 •

• -No melting; f)-partial melting; O -complete melting; ... -chemically analyzed value of mechanically separated phases in

two liquid regions.

12

l

I f

I

I ~. ~

4. Results and Discussion

4.1 . Nb20 ; and 8 20 3 Components

4.2. Phase Diagram

No quenching ex perime nts were made with the co m­ponents, as they were the same materials that had been used previously in phase equilibrium s tudies origi­nating in thi s laboratory [3, 4]. As regards the Nb20 s co mponent, the same remarks pertain as for the Nb20; - Ge02 sys te m [1]. For purposes of consiste ncy with previous publications the indicated melting te m­perature of B20 3 is given as 450°C, although this value may be 5° to 10° low [5].

The phase diagram for the system is shown in fi gure 1, and table 1 lists the compositions studied, heat treatments, and phases identifi ed. The system is characterised by a large region of liquid immiscibility , ex te nding from about 10 mole percent to 65.7 mole percent Nb20 s, and by a binary compound, "3Nb20 s . B2 0 /' melting incongruently at 1150 °C, some 200° below the monotectic.

The two triangles in figure 1 designate the c hemi­cally analyzed values for the B20 3-rich and the Nb20 s-

TABLE 1. Experimental data/or compositions in the binary system Nb2 0S - B20 3

Composition I Heal trea tm e nt 'l Result s NO les

NbzOh 820 3 Temp. T ime Ph ys ica l observa tion X-ray diffrac tion analyses 3

MoLe % Mole % 'C fl our 4.4 95.6 900 3 Considerable melling. 3, I + glass hump + [1-bBO:d

1061 I Con side rable melting. 11 09 1.5 Considerable melt ing. 3 , I + glass hurnp + [L -Nb:.!o.~J No 11 - Nb20 5 detected . 1175 I Conside ra ble me lli ng. 3: 1 + glass hurnp + [L - Nb:!0 51 1216 I Conside ra ble me lting. [H, BO,d + 3, I + [L-Nb,O, [+ H-Nb, O, Tube leaked. 1235 2 Conside rable me lti ng. [L - Nb,0 , j +3, I + [I·I,.BO,.[ + H - Nb,O, 1246 2 Ncar comp lete me lting. 3, I + [Ii , BO,]+ [L-Nb, O, ]+ H-Nb,O, 1256 2 Co mple te me lting. [L-Nb,0 ,] + [3, 1[ 1265 2 Co mple te me llin g. [L-Nb,0 , [+ [3, 1[

9.7 90.3 900 3 So me me lt ing. 3: I + g:lass hump 11 61 68 Considera ble me lting. 3, I + H -Nb,O, Aboye decomp. 3: I . 1265 2 Conside rable mel ting_ [L-Nb,0,] +3, I Ve ry wea k x- ra y patt e rn. 1310 1.5 Conside ra ble me lt ing. 3, I + IL -Nb,O, j + H-Nb,O, 1331 1.5 Con side rable me lting. 3, 1 + [L -Nb,O,[ + H-Nb,O, 1339 1.5 Cu nside rable melt ing. 3, 1 + [L - Nb,0 , [+ [l1 ,, 1l0"[ Tube leaked. 1345 1.5 Ncar cumplete me lting. [L-Nb,O, [+ 3, I + II - Nb,O,. 1350 1.5 Co mple te melting. [L-Nb,O,] One li{luid.

27.2 72.8 915 4 S li ght me lting. 3, I B2;O:I- ri ch glass present. 1180 53.5 Some melt in g_ 3, I + H-Nb,O, Above deeomp. 3: I. 1200 66.5 So me rn e lti"g. 3, I + H-Nb,O, 1275 15 Some me lting_ 3, I + H-Nb,O, 1300 18 Some mdting. H-Nb,O,+3, I + [L -Nb, O, [ 1300 18) 3, 1+ H - Nb,O, {ShOWS revers ibilit y: 11 25 11 7 3, 1- H-Nb,O,+ liqllid . 1349 2 So me melting. H -Nb,O.,+ [L -Nb,O, j + 3, I 1354 1.5 Comple te me lting. [L - Nb,O,. j + [Ii - Nb,O.,[ Above monotcc ti c. 1357 I Co mple te me lting. Chem. Ana lys is of each phase. 1380 2 Comple te me ll ing.

50.2 49.8 91 5 4 Sl ight me ltin g. 3, I 1140 69 Sl ight me ltin g. 3, 1+ IH"BO"] 1150 45 So me me it ing. 3 , I 1210 40 Some me lt ing. 3, 1+ H -Nb,O, Above decom p. 3: I. 1275 15 So me mel ti ng. Ii -Nb,0 ,+3, I 1300 18 Conside rable me lting. H-Nb,O,+3, I 1300 18) 3, I + H-Nb,O, {S hOWS reve rs ibilit y:. . 11 25 11 7 1325 I Cons ide rable melting. 3, I + H-Nb,O,

3: I ~ H -Nb20 s + li q Ui d.

1345 1.25 Conside rable meltin g. 3, 1+ H-Nb,O, 1350 I Conside rable melting. H - Nb,O, + [L -Nb,O,] + 3, 1 Decomp. 3: 1 s luggis h. 1356 1.5 Co mplete melting. [11 -Nb,O,j + IL -Nb,O, ] Above monotecti c. 1360 1.25 Comple te melting. for chem. analysis. 1380 2 Complete me lting. Above monolectic.

67.8 32.2 915 4 S light me ltin g. 3, I B2 0 :,-rich glass present. 1140 69 S light me ltin g. 3, I 1150 45 S light meltin g. 3, I 1170 20 Melting. 3, I + H - Nb,O,(?) 1180 53.5 Melting. 3, I + H- Nb,O, 1194 1.75 Melting. 3, I Decomp. 3: ] s luggish. 1200 66.5 Me lt ing. 3, I + H-Nb,O, 1220 8.5 Me lting. 3, 1+ H-Nb,O,. 1240 1.5 Melting. 3, I + H-Nb,O., 1275 1.5 Melting. H-Nb,O.,+3, 1 1300 18 Me lt ing. I1 -Nb,0, + 3, I (?) 1300 18} 3, I + Ii -Nb,O, {ShOWS reve rs ibi lit y: 11 25 11 7 3 : I ~ H - Nb20 s + liquid . 1325 I 3, 1+ H-Nb,O, 1350 I H-Nb,O,+ [L-Nb,O,j Above monot ectic. 1356 I Considerable melting. H-Nb,O, 1361 1.5 Near complete melting. H-Nb,0,+ [3,1[ 1367 I Comple te melting. [L -Nb,O,]+ [3,1[ One liquid . 1380 2 Comple te mehing. [L-Nb,0,]+13, I]

13

TABLE 1. Experimental data/or compositions in the binary system Nb20 S - B,03 - Continued

Composition 1 H eal t reatment 2 Result s

Nb,O, B20 5 Te mp . Time Ph ys ical observation X.ra y diffrac tion analyses 3 Notes

Mole % Mole % °C Hour

73 .0 27.0 900 3 Opalescent. 3: I 11 35 336 Opalescent. 3 : I 1140 69 Opalescent. 3 : I 1161 68 Opalescent. 3: I + H-Nb,O, Above decornp. 3: 1. 1300 18 Opalescent . H-Nb,O, + 3 Ii ?) 1300 18) H- Nb,O,+3 : 1 {S hows reve rsibilit y: 1125 117 3: I ~ H - Nb20 !> + liquid. 135 1 1.5 S light me lt ing. H-Nb,O, 1355 1.5 Melting. H - Nb,O,+ 3: 1 Ahov!:: mon oleclic. 1380 1.5 Cons ide rable melting. H-Nb,O,+ 3: I 1385 1.5 Near comple te melting. H -Nb,O,+ [L - Nb,O,] + 3 I 1390 1.5 Near complet e melting. [L - Nb,O,]+ H-Nb,O., + 3: I 1398 1.5 Complete me lting. [H -Nb,O,]+ [L -Nb,O,]+ [3· I] All dev itrify from liquid.

77.8 22.2 900 3 No me lting. 3: I 11 35 336 No me lting. 3 : 1 1140 69 "Io melting. 3 : 1 1161 68 Slight me lting. 3: I + H- Nb, O, Above decomp. 3 : l. 1300 18 Opalescent . H-Nb,O:. + 3 : I (?) 1300 18) H-Nb,O,+3 : I {S hOws re ve rs ibilit y: 11 25 11 7 3 :1 ++ H-Nb20 5 + liquid. 1401 1 Cons ide rable melting. H -Nb,O,+ [L- Nb,O,] + 3 I 1409 2.5 Near comple te melting. H- Nb, O, + [L -Nb,O, I+ 3 : I 141 5 1.5 Comple te me lting. [H -Nl>, O, I+ [1. - Nb,O, 1 + 13 : 11 1420 1 Complete me lting. fll - Nb, O, 1 + [L - Nb,O,1 1440 1 Complete me lting. [L - Nb,O,1 + I H - Nb,O.,1

82.9 17. 1 900 3 No me lting. 3: I 11 35 336 No me lting. 3 : I + H-Nb,O, 1161 68 S light me lt ing. 3: 1 + H -Nb,O" Above d cco mr). 3 : l. 1300 18 Melting. H - Nb,O, + 3 : 1 (?) 1300) 18 H-Nb,O,+ 3: 1 Re ve rsibilit y sluggish. 1125 11 7 1349 3 Melting. H - Nb,O, + I L - Nb, O,] Below monotec ti c. 1359 3 Considerable me lting. H - Nb,O:. + [L - Nb, O.,] Above monotecti c . 1420 I Considerable melting. H -Nb,O, + [L - Nb, O, I+ 3 I 1425 1.5 Nea r c umple te melting. H-Nb,O,+3 : 1 (?) 1430 1 AI liquidus. [H - Nb,O,I+ [L - Nb,O, 1 + 13: 1 (?) I 1440 1 Compl ete me lting. [H - Nb,O, I+ [L -Nb,O, 1

87.2 12.8 900 3 No me lting. 3: I + H - Nb,O, (I) 11 35 336 No me lting. 3 : 1 + H- Nb,O, 1161 68 Meltin g. 3 : I + H-Nb,O, Above deco mp. 3: 1. 1300 18 Melting. H-Nb,O,+ 3: 1 (?) 1348 2 Mellin g. H-Nb,O, NIII < 1.47 , b elow monotec li c . 1356 2 Mode rate me lting. H-Nb,O, Above monotecti c. 1358 1.5 Mode rat e me lting. H- Nb,O, 1362 1.5 Moderate me lting. H-Nb,O:. 1440 I Considerable me lting. H -Nb,O, + 3: I (?) 1445 I Nea r comple te melting. H -Nb,O, + [L -Nb,O,]+ [3 : I] 1450 I Complet e me lting. IL-Nb,O,]+ [H - Nb,O,]

L Analyzed value fo r Nb20 5 : B20 3, by diffe re nce. 2 S pec;me ns quenc hed in sealed Pt tubes. 3 Phases ide ntifi ed are li s ted in orde r of amount present at room te mpe rature. Phases not necessarily present at the e levat ed te mperatures a re

e nclosed in bracke ts. 3: 1 re fers to 3N b20 s · B20 :\, the most likc ly composition of the binary compound .

rich devitrified liquids which were mechanically separated from samples quenched from above the monotectic. When the Pt was removed from the quenched specimens, two distinct layers were clearly visible. The top layer appeared opalescent white and was found to be hard and glassy when ground. An index of refraction of approximately 1.51 ± 0.01 was determined for clear portions of some of the grains. The lower layer appeared crystalline, metallic grey in color. Under the microscope the grains were dark and very weakly birefringent. The x-ray powder patterns showed that the top layer (B20 3-rich) con· tained low-Nb20 5 with a trace of high-Nb20 5 , whereas

14

the lower layer (Nb20 5-rich) high-Nb20 5 , low-Nb20 5, and "3Nb20 5 ' B20 3."

contained appreciable a small amount of

Homogeneous glasses were never obtained for any composition as the liquids devitrified on quenching. X-ray examination showed either low-Nb20 5 , high­Nb20 5 , "3Nb20 5 ' B20 3", or combinations. The low­Nb20 5 phase corresponded to the pseudo-orthorhombic subcell of low-Nb20 5 found for some of the quenched liquids in the Nb20 5 - Ge02 system. It had the same unit cell dimensions [1]. The metastable low-tempera­ture niobia-type phase of hexagonal symmetry oc­curring in the Ge02 system was not found.

i I

\

4.3. Compound "3Nb20 5 • 820 3"

The exac t composition of thi s compound could not be determined because of insufficient sensitivity, in thi s in stance, of the microscopic and x-ray me thods toge ther with some uncertainty in the chemical anal­yses. Therefore, the 3: 1 ratio of oxides is given as the most probable.

The co mpound formed readily for all compositions in the system. It melted incongruently at about 1150 °C to liquid and H-Nb20 5 • However, disappearance of the compound above the melting point was ex­tre mely sluggish. The compound could b~ detect~d in diminis hin a amounts up to the monotectIc and lIq­uidus temper:tures (see table 1). Even compositions heated at 1161 °C for 68 hr showed appreciable amounts of the compound, but the appearance of H - Nb20 5 indicated that the temperature was abo~e the incongruent melting point. X-ray powder dIf­fraction data suitable for identification is given in table 2. The pattern is of low symmetry with regions of closely spaced low-intensity peaks and could not be indexed.

TABLE 2. X-ray powder diffraction data (( uK" radiation) fo r the compound "3Nb,O, . 820 ; '

(P atte rn of low symme try a nd could not be indexed)

d I II , d 1/10

A % A % ' 14.4(b) 4 2.434 3

7. 16 7 2.389 6 6.93 6 2.369 4 4.774 34 2.337 3 4.741 27 2.302 17 3.770 71 ' 2.094(b) 3 3.584 90 2.049 26 3.553 100 2.033 27 3. 170 12 1.91 25 33 2.864 6 ' 1.79 18(b) 5 2.833 10 1. 7746 10 2.773 57 1.7697 10

'2.730(b) 7 1.6995 22 2.675 9 1.6874 21 2.576 6 1.6842 23 2.520 7 ' 1.5838(b) 6 2.506 9 1.5769 16 2.473 4

*b = broad peak.

4.4. Application to Liquid Immiscibility Theory

The niobium borate and niobium silicate systems both contain large regions of immiscibility, whereas the niobium germanate system shows complete liquid miscibility. Thus, as previously discussed [1], the niobium cation possesses a critical ionic field strength for the formation of two liquids with the glass-forming cations.

With data on immiscibility in the niobium borate and silicate sys tems, it is possible to test and to extend the principles of immiscibility developed mostly from data on the divale nt cations to the cations of high c harge . A fundamental consideration in the struc­tural approach is the average number of oxygens associated with a modifier cation, in the modifier-rich

15

liquid. Once this value is known, statement of the nominal composition is a routine calculation. The oxygen-volume method, discussed in previous publica­tions [3, 6] is an attempt to make a reasonable estimate of thi s number.

TABLE 3. Calculated composition of modifier-rich liquids, in #O/Nb and in mole percent Nb20 5 , based on the oxygen-volume method [6] and two values both for the niobium separation and the volume of space occupied by 0-' (Vo).

Calcul ated from Nb-N b separation:

Experimental of 4.18 ).[2(1.40 + 0.69)], and . of 4.08 A, and System

Vo = 17 A' Vo= 19 A' Vo= 19 A'

Nb,O, #O/Nb #O/N b Nb,O, # O/Nb Nb,O, #O/Nb Nb,O,

MoLe % Mole % Mole % Mole % Nb,O, - B,O" 65.7 3.28 4.1 8 47.2 3.72 55. 1 3.45 61.1 Nb,O, - 5;0 , [21 47.5 3.61 4 .18 37,4 3.72 45.0 3.45 51.2

Table 3, column 3 gives the number of oxygens per modifier cation calculated directly from the experi­mentally determined composition of the modifi er-ri ch liquid, For example, in the Nb20 5 -B20 3 sys tem:

/Nb = 5 X 0, 657 + 3(1.00 - 0,657) = 3,28, #0 2 X 0,657

_ 5 X 0,475 + 2(1.00 - 0,475) = 3 61 #O/Nb - 2 X 0,475 ' ,

As a first approximation the number of oxygens per cation may be co nsidered independe nt of the glass­formin g cation, Using the data in column 3 for one of the systems, it is possible to calculate the compo­sition of the modifier-ric h liquid in the other sys tem. The calculation yields 57.5 mole percent Nb20 5 in the borate system,2 using the 3.61 value, and 56.2 mole percent Nb20 5 in the silicate system, using the 3.28 value. The calculated value in the borate sys­tem is about 8 mole percent low, that in the silicate system about 9 mole percent high. As will be shown later, this seemingly large discrepancy is due to the sensitivity of the calculations in the niobium oxide systems.

Table 3 (columns 4-7) lists the calculated #O/Nb and the corresponding mole percent Nb20 5 , based on the oxygen-volume method using a Nb - 0 - Nb bond angle of 180° and Ahrens' ionic radii for Nb+5 and 0 -2. In the original papers on immiscibility the value of 17 A.3 was taken as the average volume of space occupied by an oxygen atom in the modifier-rich liquid.

2 Nominal oxide formu la Nb,O; . (2 x 3,61- 5) B,OJ = Nb,O,' 0.74B,O, (57.5 mole %

Nb,O,).

An analysis of the experimental data in about 40 borate and silicate systems showing iJVmiscibility [7] gives an empirical value of about 19 N. There­fore, in table 3 calculations are given for bpth oxygen values. It may be seen that for the 19 N value in the silica case agreement is good between the calcu­lated mole percent Nb20 5 (45.0) and the experimental one (47.5). In the borate case the calculated value is 11 mole percent low.

For the case of small highly charged modifier ca­tions, such as niobium, the calculations are very sensitive to cationic radius because a small difference in separation has a large effect on the calculated number of oxygen per cation. For example, using a separation of only 0.1 A less than 4.18, will increase the calculated value by about 6 mole percent (column 9); and the calculated number of oxygens per niobium (column 8) is intermediate between 3.28 and 3.61, the experimentally determined values for the borate and silicate systems, respectively. Gatehouse and Wadsley [8], from a crystal structure determination of high-Nb20 5 , have concluded that in Nb20 5 the average ionic radius of 0 2 - is L.40 A and that of Nb5+, 0.59 A. The latter value is 0.1 A less than Ahrens' ionic radius.

5. Summary

The phase equilibrium diagram for the system Nb20 5 - B20 3 was constructed from "quenching" data on nine selected compositions. Solidus and liquidus values were determined by examination of the samples with the binocular and polarizing micro­scopes and x-ray powder diffractometry.

The system was found to contain one binary com­pound, "3Nb20 o ' B20:t, which melts incongruently at about 1150 °C, some 200 below a monotectic. The liquidus is characterized by a large two-liquid region, at 1352 °C, extending from about 10 mole percent Nb20 5 to 65.7 mole percent Nb20 5•

16

Experimental limitations precluded an unequivocal statement of the composition of the "3Nb20 5 • B20 3 "

compound. Above the incongruent melting point its decomposition to H - Nb20 5 and B20 3-rich liquid is sluggish. X-ray powder data suitable for identifica­tion is listed.

In the modifier-rich liquid in the immiscibility region, an average of 3.28 oxygen are associated with each Nb atom, as compared to 3.61 for the corresponding silicate system. Calculation of the composition of the modifier-rich liquid, by the oxygen-volume method, gives acceptable agreement with the experimentally determined value, considering that for cations of large valence the calculation is sensitive to small changes in the parameters.

6. References

[1] E. M. Levin , J. Res. NBS 70A (phys. and Chem.), No.1, 5-10 (1966).

[2] M. Ibrahim and N. F. H. Bright, J. Am. Ceram. Soc. 45 [5], 221-22 (1962).

[3] (a) E. M. Levin, C. R. Robbins, and J. L. Waring, J. Am. Ceram. Soc. 44 [2], 87- 91 (1961).

(b) E. M. Levin and C. L. McDaniel , J. Am. Ceram. Soc. 45 [8J, 355- 60 (1962).

[4] (a) 1. L. Waring and R. S. Roth, J. Res. NBS 69A (phys. and Chern.), No.2, 119-29 (1965). .

(b) R. S. Roth and 1. L. Waring,1. Res. NBS 65A (phys. and Chern.), No.4, 337- 44 (1961).

[5] T. 1. Rockett and W. R. Foster, J. Am. Ceram. Soc. 48 [2], 75-80 (1965).

[6] E. M. Levin and S. Block, J. Am. Ceram. Soc. 40 [3], 95-106; [4], 113- 18 (1957).

[7] E. M. Levin, Structural interpretation of immiscibility: parts IV, V, and VI, in preparation.

[8] B. M. Gatehouse and A. D. Wadsley, Acta erys!. 17 [12], 1545--54 (1964).

(Paper 70AI-382)

A I